1 Introduction

Since the 2000s, a lot of football specialists or just football lovers are disappointed about football. Indeed, money tend to parasite football and to have more and more influence in the football world. A lot of teams have been purchased by billionaires who come often from foreigners’ countries like Qatar, United Arab Emirates, Russia or the USA. They want results immediately and they are not afraid to spend a lot of money. Other teams are aware of that and ask for more money to sell their players. Indeed, the price of player-transfers are rising inexorably, as the player salaries.
This is not the only explication, the increase in the amount of television rights is another variable. The boom in these rights, advertising revenues, merchandising are indicative of the globalization of the football economy and its constant expansion. Football is not just a sport anymore but a real business where a lot of money is put at stake.

(For more details William 2017).

There is no more secret to be successful, you need money to be able to build a competitive team or just to keep your best players.

But what is the real influence of money in the 5 mains football leagues? Do you really need to have the more expensive team to be champion of the league? And conversely it is always the poorest who descend in second league? How evolved the disparities across time?

This is as much as questions we are going to try to answer.

1.1 Why this subject ?

Passionates about football since 2006 and the dramatic World Cup final lost by the french against the italians, we wanted to combine utility and pleasure. We tried to find a subject related to Football but this was not an easy task. After some days of reflection, we have found our subject: the influence of money in football.

A topic of conversation that every person who are passionate about football have already had. Everybody agreed to say that money has an impact, but some people say that it is a limited one, indeed a good game strategy could be sufficient to compensate an inferior amount of money to win against the richest team. We wanted to close the debate with some real statistics and not with a lot of “bla-bla” which come from nowhere.

1.2 Description of the data

In order to be able to analyse the impact of money in football, we needed on one hand to find data representing sports results and on the other hand looking for financial data for each teams.

For the financial part we have had to modify a scrap script. Our goal was to have the mean player market value for every team in the 5 biggest championships and for the last 10 years. That means from 2010 to 2019.

These leagues are:

  • “Premier League” for United-Kingdom championship

  • “La Liga” for Spain championship

  • “Serie A” for Italy championship

  • “Bundesliga” for Germany championship

  • “Ligue 1” for France championship

Unfortunetly we didn’t found any table online for the ranking so we have choosen to build this table on our own based on others tables found on this link. We have then extract 50 tables from this website summarizing the results of each match in the 5 championships during 10 years. With these tables we have built a function for to have the ranking at the end of the season. After that we have renamed around 200 teams names manually to match our new ranking table with the market value table built with scrap_transfertmarkt.R Finally after left_joining both tables we have our mw_table on which we could start the work.

You can find our scrap_transfertmarkt.R and our table-building.R scripts in the Table_building file in our project. Furthermore, we have decided to put the mw_table directly in our Data file for practical reasons but it can be found by just running the table-building.R script.

2 Exploratory Data Analysis

First of all, it is important to know that a team with a high mean market value per player is a team which had a lot of money. Indeed, players with high market value have higher salaries and by definition are more expensive to buy/sell.

For us the mean market value per player will be an indicator of wealth.

2.1 A high evolution in term of player market value

Map showing the mean market value evolution

Figure 2.1: Map showing the mean market value evolution

The figure 2.1 shows us a clear evolution of the market value per player in the five main leagues since 2010.

England is far ahead since the 2010. Indeed, the Premier League has the highest rise with an increase of mean market value of approximately 5 million euros in 10 years. La Liga and the Bundesligua have a slower evolution respectively from 4 to 7 million and 3 to 6 million. The Ligue 1 was lagging behind the mean market value per player from the four other countries, but in 10 years made a nice progress and almost catch up Italy, which had the smallest growth of only 1.5 million in 10 years.

Whatever the evolution, football seems to have become more important in each country.

Evolution of the mean market value per player

Figure 2.2: Evolution of the mean market value per player

Here we will take in considerations teams which were present in their league since 2010.

The graph 2.2 shows us their evolution in term of mean market value per player.

We can clearly see a rise in each league. As seen before 2.1, the Serie A have the smallest evolution. Apart from Juventus and on a smaller scale, Napoli and Milan AC, the mean market value per players from all the others teams seem to stagnate.

The Bundesliga and the Ligue 1 have a similar evolution with one team increasing largely compared to the others. FC Bayern Munchen in Germany and Paris Saint-Germain in France with respectively an evolution going from 10 to 25 million and from 4 to 21 million in ten years. Few teams from both leagues increase on a smaller scale like BVB Dortmund, Bayer Leverkusen or Olymplique Lyonnais.
It is almost the same case for Spain, but with 2 teams which stand out, FC Barcelona and Real Madrid FC with an evolution of 10 million in 10 years. Since 2017 an other team stand out: Athletico Madrid with an evolution of 10 million in the 3 last years. All other teams have at least a small evolution.

All the observed teams in the Premier League have a high and tight evolution. Only Everton FC is trailing behind. In contrary to others league, in England, it is a group evolution. Five teams exceed the 18 million mean market value per player.

2.2 The power of money

Boxplot showing the importance of money

Figure 2.3: Boxplot showing the importance of money

Regardless of the league, in general, it is the team with the highest market value who is the champion at the end ( 2.3). The three firsts teams have largely a higher market value. Specially the first one with a median of 13 million, corresponding to 4 million more than the second one. Until rank 9, the ranking is consistent with the market value. After these ranks, we can notice that market value seems to have less impact. However, the difference in the market value is less important. From place 10 to 20, the difference in the median for each team is less than 1 million and 4 million starting from rank 4.

Detailing graph from the rankings

Figure 2.4: Detailing graph from the rankings

Let’s look more in detail at the influence of money 2.4. We can clearly see that the ranking increase with the mean market value per player, better your mean market value per player is and better your ranking would be.

In majority the overall winner (point in red) is the one with the highest market value (more on the right part of each x-axis). In contrary to the last one (point in orange) which is rarely the one with the less market value.

In Ligue 1, Serie A and Bundesliga, it is often the team with the highest mean player market value who win, with respectively 7, 9 and 8 times out of ten. In Spain it is a little bit better, since 2010, the champion was 6 times the team with the highest market value. In England, it is only 4 times the case. As seen before 2.2 it is the league where there was a grouped evolution in term of mean market value per player. This is the league the most competitive and that is why it is even more complicate to be champion even if you have the most money.

Concerning the last place in Spain and England, since 2010 it is never the team with the poorest mean market value per player which was the last of the ranking. It was only the case 2 times in Italy and France, but 6 times in Germany. It is because there are no big differences between the last ranks in term of market value 2.3.

Graph showing leagues ultra dominated

Figure 2.5: Graph showing leagues ultra dominated

As seen before 2.4 , the champions are often the teams with the highest mean market value per player.

As shown in 2.2, in the majority of the leagues, one or two teams dominate the others in terms of market value. This differentiation is visible in term of ranking 2.5.

Indeed FC Bayern Munich , Paris SG and Juventus which clearly dominated their leagues in terms of market value, were champions respectively 8 times, 6 times and 8 times out of 10. Concerning La Liga, Real Madrid FC and FC Barcelona shared the first place for 9 years with 3 titles for Real Madrid and 6 for FC Barcelona. There is not a lot of suspense regarding these leagues 2.5.

The Premier League is the only league where there is not a clear domination.

Graph showing a tight championship

Figure 2.6: Graph showing a tight championship

In England 2.6 a lot of teams have high market value and contrary as other leagues it is not just one or two teams. It is a very competitive league where a lot of teams can win and not necessarily the richest one. The best example is Leicester which wins the league in 2016 with the 4th lowest market value per player.

A graph of the evolution of the market value per ranking in each league

Figure 2.7: A graph of the evolution of the market value per ranking in each league

The graph 2.7 justify what we have seen above 2.5. Indeed, the difference in term of market value between the first teams and the others is less important in England than in other countries. Moreover, the Premier League is the only league where the firsts 6 teams have a so small difference in term of market value, approximately 4 million. In contrary to other leagues, there is a smaller disparity and more surprise can appear as shown in 2.6. That is why it is the most attractive league.

2.3 Disparities in Football

Graphs showing various growing disparities about rankings

Figure 2.8: Graphs showing various growing disparities about rankings

These graphs 2.8 show us the disparities which appears since 2010.

First regarding the number of points, the difference between the first and the last rank had increase a lot. Then concerning the mean market value per player, the firsts places have a faster increase than the lasts ones in term of mean market value per players.

The best teams (with more money) have more points and vise versa. However, there is no much change in the number of points for teams in the middle of the ranking. This can be explained by the fact that now they are not able anymore to win against the richest teams but the poorest teams are not able to beat them. All results are more predictable and football become less interesting.

2.4 The Champions League

Graph representing the winners of the Champions league

Figure 2.9: Graph representing the winners of the Champions league

The champions league is a competition regrouping the best teams of each league in Europe. Which means the richest teams with the highest market value of Europe. It is a competition comparable to the Premier League (2.6), where there are a lot of teams with high market value. Money have less impact because you play against teams which had approximately the same market value. That creates a more competitive competition, very difficult to win. Indeed, during 10 years, teams with the highest market value won only 4 times (2.9).

3 Modeling

In this section we will try to understand the impact of the market value on the final ranking by first, building a model and have a look at all the parameters. Then we will assess the goodness of fit and see if we can predict some good results.

Let’s start by plotting a scatterplot of the number of points regarding the market values:

Distribution

Figure 3.1: Distribution

We can see in 3.1 graph that the number of points increase exponentially with the market value. In order to have better results we will apply the logarithm function 3.2 on the market value to have linear distribution and then use a linear model.

Log Distribution

Figure 3.2: Log Distribution

3.1 Model

Our model is based on the following formula * points ~ log(mw) + name + year. Furthermore we will use the linear least square method to fit all the parameters.

Table 3.1: Summary of the model for Premier League
Parameter Estimate Standard Error P-value Significance
(Intercept) 969.976 258.553 0.000 ***
logmw 5.791 1.111 0.000 ***
year -0.464 0.128 0.000 ***
nameAjaccio GFCO 7.285 10.805 0.500
nameAlaves 14.253 7.638 0.062 .
nameAlmeria -2.145 7.134 0.764
nameAmiens 7.899 8.529 0.355
nameAngers 12.050 7.136 0.092 .
nameArles -14.921 10.760 0.166
nameArsenal 23.908 6.619 0.000 ***
nameAston Villa -2.775 6.609 0.675
nameAtalanta 12.863 6.250 0.040
nameAth Bilbao 12.240 6.320 0.053 .
nameAth Madrid 24.984 6.592 0.000 ***
nameAugsburg 3.216 6.325 0.611
nameAuxerre 12.856 7.628 0.092 .
nameBarcelona 40.668 6.956 0.000 ***
nameBari -0.648 8.516 0.939
nameBastia 10.746 6.809 0.115
nameBayern Munich 29.351 6.827 0.000 ***
nameBenevento -13.720 10.772 0.203
nameBetis 6.949 6.486 0.284
nameBirmingham 2.926 8.580 0.733
nameBlackburn -0.597 7.702 0.938
nameBlackpool 3.206 10.759 0.766
nameBochum -10.866 10.777 0.314
nameBologna 4.107 6.234 0.510
nameBolton -1.891 7.711 0.806
nameBordeaux 14.891 6.217 0.017
nameBoulogne -5.219 10.762 0.628
nameBournemouth 4.989 7.193 0.488
nameBraunschweig -9.651 10.757 0.370
nameBrescia -5.254 10.763 0.626
nameBrest 2.530 7.607 0.740
nameBrighton -2.094 8.601 0.808
nameBurnley 0.403 6.858 0.953
nameCaen 7.429 6.435 0.249
nameCagliari 3.615 6.247 0.563
nameCardiff -6.684 8.548 0.435
nameCarpi 1.104 10.770 0.918
nameCatania 6.666 6.832 0.329
nameCelta 8.230 6.494 0.205
nameCesena -7.319 7.612 0.337
nameChelsea 24.813 6.788 0.000 ***
nameChievo 5.752 6.136 0.349
nameCordoba -15.884 10.760 0.140
nameCrotone 2.395 8.532 0.779
nameCrystal Palace 3.678 6.716 0.584
nameDarmstadt -0.741 8.526 0.931
nameDijon 5.093 7.129 0.475
nameDortmund 20.444 6.514 0.002 **
nameEibar 10.311 6.822 0.131
nameEin Frankfurt 4.221 6.266 0.501
nameElche 4.194 8.509 0.622
nameEmpoli 5.391 7.129 0.450
nameEspanol 8.172 6.189 0.187
nameEverton 11.027 6.447 0.088 .
nameEvian Thonon Gaillard 6.722 7.116 0.345
nameFC Koln -1.497 6.484 0.817
nameFiorentina 13.016 6.345 0.041
nameFortuna Dusseldorf 2.070 8.511 0.808
nameFreiburg 2.260 6.233 0.717
nameFrosinone -6.989 8.520 0.412
nameFulham -1.507 6.744 0.823
nameGenoa 4.492 6.211 0.470
nameGetafe 8.143 6.258 0.194
nameGirona 7.912 8.535 0.354
nameGranada -2.330 6.629 0.725
nameGrenoble -14.297 10.765 0.185
nameGreuther Furth -14.090 10.756 0.191
nameGuingamp 8.823 6.600 0.182
nameHamburg -2.351 6.333 0.711
nameHannover -0.348 6.262 0.956
nameHercules -5.273 10.794 0.625
nameHertha 0.197 6.367 0.975
nameHoffenheim 4.034 6.239 0.518
nameHuddersfield -11.848 8.564 0.167
nameHuesca 0.359 10.783 0.973
nameHull -6.702 7.195 0.352
nameIngolstadt 2.155 8.515 0.800
nameInter 19.067 6.512 0.004 **
nameJuventus 35.988 6.620 0.000 ***
nameKaiserslautern -3.339 8.516 0.695
nameLa Coruna -0.945 6.475 0.884
nameLas Palmas -1.758 7.632 0.818
nameLazio 19.535 6.277 0.002 **
nameLe Mans -6.611 10.774 0.540
nameLecce 2.654 8.505 0.755
nameLeganes 4.681 7.637 0.540
nameLeicester 10.876 6.972 0.119
nameLens 1.853 7.607 0.808
nameLevante 6.982 6.323 0.270
nameLeverkusen 13.285 6.420 0.039
nameLille 22.456 6.218 0.000 ***
nameLiverpool 20.529 6.639 0.002 **
nameLivorno -10.482 8.512 0.219
nameLorient 9.201 6.320 0.146
nameLyon 24.609 6.365 0.000 ***
nameM’gladbach 8.672 6.271 0.167
nameMainz 6.270 6.165 0.309
nameMalaga 5.060 6.285 0.421
nameMallorca 8.417 7.168 0.241
nameMan City 31.223 6.771 0.000 ***
nameMan United 26.367 6.735 0.000 ***
nameMarseille 20.298 6.366 0.001 **
nameMetz 0.670 7.631 0.930
nameMiddlesbrough -12.931 10.835 0.233
nameMilan 20.270 6.499 0.002 **
nameMonaco 22.601 6.504 0.001 ***
nameMontpellier 17.538 6.147 0.004 **
nameNancy 5.581 6.812 0.413
nameNantes 13.236 6.602 0.045
nameNapoli 29.276 6.450 0.000 ***
nameNewcastle 2.648 6.499 0.684
nameNice 15.485 6.165 0.012
nameNimes 18.973 10.779 0.079 .
nameNorwich 0.516 7.152 0.942
nameNovara -3.157 10.757 0.769
nameNurnberg -2.588 6.599 0.695
nameOsasuna 2.551 6.608 0.700
namePaderborn -2.360 10.763 0.826
namePalermo 5.712 6.460 0.377
nameParis SG 32.850 6.616 0.000 ***
nameParma 8.393 6.458 0.194
namePescara -13.924 8.509 0.102
namePortsmouth -13.578 10.817 0.210
nameQPR -11.281 7.713 0.144
nameRB Leipzig 19.223 7.797 0.014
nameReading -9.597 10.765 0.373
nameReal Madrid 36.060 6.930 0.000 ***
nameReims 10.684 6.809 0.117
nameRennes 12.396 6.205 0.046
nameRoma 27.675 6.421 0.000 ***
nameSampdoria 9.201 6.285 0.144
nameSantander -2.668 7.654 0.727
nameSassuolo 7.513 6.640 0.258
nameSchalke 04 8.267 6.394 0.196
nameSevilla 15.655 6.394 0.015
nameSiena 0.181 7.611 0.981
nameSochaux 6.237 6.822 0.361
nameSociedad 10.784 6.355 0.090 .
nameSouthampton 5.659 6.632 0.394
nameSp Gijon 1.354 6.815 0.843
nameSpal 5.403 8.528 0.527
nameSt Etienne 17.496 6.211 0.005 **
nameSt Pauli -7.478 10.760 0.487
nameStoke 2.903 6.386 0.649
nameStrasbourg 10.056 8.529 0.239
nameStuttgart -0.203 6.338 0.974
nameSunderland -3.306 6.461 0.609
nameSwansea 3.313 6.545 0.613
nameTenerife -0.349 10.762 0.974
nameTorino 13.740 6.479 0.034
nameTottenham 23.200 6.605 0.000 ***
nameToulouse 6.799 6.172 0.271
nameTroyes -3.297 7.624 0.665
nameUdinese 9.559 6.198 0.123
nameValencia 15.977 6.474 0.014
nameValenciennes 5.939 6.821 0.384
nameValladolid 2.744 7.119 0.700
nameVallecano 6.320 6.598 0.338
nameVerona 2.948 7.122 0.679
nameVillarreal 14.306 6.391 0.025
nameWatford 2.652 7.255 0.715
nameWerder Bremen 3.727 6.210 0.549
nameWest Brom 1.313 6.427 0.838
nameWest Ham 1.444 6.411 0.822
nameWigan -1.638 7.188 0.820
nameWolfsburg 3.589 6.361 0.573
nameWolves -0.372 7.180 0.959
nameXerez -4.251 10.771 0.693
nameZaragoza 0.294 7.177 0.967
3.1 is the list of the parameters and their estimates. We can see that the logmw increase the number of point and it’s a very good point for our assumption. The intercept is very high but it balanced the year coefficent which is -0.464 (*2010 for instance). Furthermore we can see that only biggest teams have a significant estimate as Man City, Real Madrid, Barcelona etc.. with very high value, around 30. There is of course an impact of the team on the ranking. All others teams have a value which are not very significant, it could be the case because their market values are quite all the same under the 10th rank and their number of points are then more difficult to differentiate and predict.

Here 3.2 we have compared the Rsquared adjusted of two models:

  • points ~ log(mw) + name + year.
  • ranking ~ log(mw) + name + year.
Table 3.2: Summary of the model with points
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual
0.738 0.682 9.31 13.3 0 172 -3483 7312 8158 70109 808
Table 3.3: Summary of the model with rank
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual
0.661 0.59 3.64 9.22 0 172 -2562 5470 6315 10701 808

We can see that the Rsquared adjusted is around 68% in the point model against 59% in the ranking model, that is why we have prefered to take the number rather than the ranking.

68% is quite good results for our small number of explained variables.

3.2 Goodness of fit

To Discuss the goodness of the fit we will plot the residuals and have a look at some predictions made by our model.

Residuals vs Log Market value

Figure 3.3: Residuals vs Log Market value

Residuals vs fitted values

Figure 3.4: Residuals vs fitted values

The residuals graphs 3.3 and 3.4 are well randomly spread with a mean of zero for the log market value and for the fitted graph. It is a good point for our model since their is no clear pattern in the residuals.

 Prediction graphs

Figure 3.5: Prediction graphs

In the last graph of our project we have built a function that takes into account the fitted number of points reach by each team and used it to build a ranking. We have then built this graph 3.5 to have a view on our prediction against the observed ranking. We have selected some years and championships to illustrate this. As we can see it predict quite well the ranking with obviously some mistakes but they are quite small in terms of ranking. In fact we have seen in the residuals graphs 3.3 that our errors are in majority under 10 points, it seems high,but in terms of ranking 10 points represent approximately 4 places depending on the season. As you can see, our errors in terms of ranking are quite small and we predict well from the 1st rank to the 10th, but after the 10th rank it is more difficult as there is not a lot of differentiation in term of market values

Our Model describes well the big influence of the market value in football. However, our model has some boundaries in our assumption. Indeed, it does not just take into account the market value but also the team… As you can imagine the estimate for Barcelona is not the same as the one for Toulouse. For instance, if tomorrow Barcelona become a low-ranking team with small market value for whatever reason, our model will still fit a high number of points because of the “background” of the club the previous years.

4 Conclusion

To conclude, we can say that the disparity in term of wealth, have killed the suspense from a lot of leagues like Spain, Germany, France or Italy. Indeed, in these leagues the richest teams have a faster evolution in terms of mean market value per player than the others and will be champions most of the time. Regarding the lasts ranks, there is more suspense because a lot of teams with small market value fights to avoid the relegation. It’s maybe here where stand the real suspense !

In contrary to most of the leagues the Premier League and the Champions league stay competitive.

References

William, Plumme. 2017. “Comment Les Prix Ont Explosé Sur Le Marché Des Transferts,” August.